batteries

Article
Electrochemical Impedance Spectroscopy and
Determination of the Internal Resistance as a Way to
Estimate Lead-Acid Batteries Condition
Włodzimierz Majchrzycki 1 , Ewa Jankowska 1 , Marek Baraniak 2 , Piotr Handzlik 3, * and
Robert Samborski 4
1 Institute of Non-Ferrous Metals, Division in Poznan Central Laboratory of Batteries and Cells, Forteczna 12,
61-362 Poznan, Poland; wlodzimierz.majchrzycki@claio.poznan.pl (W.M.);
ewa.jankowska@claio.poznan.pl (E.J.)
2 Institute of Chemistry and Technical Electrochemistry, Faculty of Chemical Technology, Poznan University
of Technology, Berdychowo 4, 60-965 Poznan, Poland; marek.baraniak@put.poznan.pl
3 Department of Physical Chemistry and Metallurgy of Non-Ferrous Metals, Faculty of Non-Ferrous Metals,
AGH University of Science and Technology, Mickiewicza 30, 30-059 Cracow, Poland
4 The National Institute of Telecommunications, Szachowa 1, 04-894 Warsaw, Poland; R.Samborski@itl.waw.pl
* Correspondence: phandzli@agh.edu.pl; Tel.: +48-12-617-41-24




Received: 1 August 2018; Accepted: 2 November 2018; Published: 12 December 2018


Abstract: Attempts have been made to find the best procedure for the detection of premature battery
capacity loss (the so called “PCL”) in AGM-VRLA 48 V batteries operating in telecommunication
systems. However, recorded changes in internal resistance and potential did not give clear indications
of the beginning of the PCL effect. The obtained correlation between internal resistance and
potential derived from used batteries does not show the expected trend in measured parameters.
It seems that the application of Electrochemical Impedance Spectroscopy (EIS), which is a faster
and non-destructive method, may solve this problem. It is demonstrated that the change in internal
resistance (which is an indicator of the state of health (SoH)) can be determined from EIS spectra
during continuous operation of 12 V monoblocks in a backup power source of a base transceiver
station (BTS).

Keywords: PCL-effect; stationary lead acid battery; state of health (SoH); internal resistance;
electrochemical impedance spectroscopy (EIS)

1. Introduction
Stationary lead-acid batteries are still the most important chemical backup power source used in
many technical applications. Therefore, problems with the appropriate level of system safety become
more and more important. Detection of increases in the internal resistance may indicate a loss of
the real battery capacity [1,2], but this test is not sufficiently reliable. Our investigation carried out
on brand new and faulty batteries showed that a battery which is not operating properly may not
show changes in its internal resistance [3,4]. It is known that a credible assessment of the battery state
consists of its discharge, but this requires that the system be turned off for several hours (like during
electrical tests such as discharge by currents lasting for 10 or 20 h).
It seems that via electrochemical impedance spectroscopy (EIS), it is more likely that faulty
batteries will be detected. The combination of the two above-mentioned techniques should allow
for the prediction and/or the identification of battery parts with significant capacity losses, without
discharging the battery; the method also shows high reliability [3–8].

Batteries 2018, 4, 70; doi:10.3390/batteries4040070 www.mdpi.com/journal/batteries

Batteries 2018, 4, 70 2 of 13

Battery capacity decrease is caused, among other factors, by: water/electrolyte loss from
absorptive glass mat (AGM) separator, grid corrosion, mass sulfation, and drop of mass [9,10]. One of
the most frequent and less known factors influencing the decrease in lead-acid battery capacity is the
Premature Capacity Loss (PCL) effect. PCL affects each type of lead-acid battery. This phenomenon
is more frequently studied in a relation to stationary batteries because of their assumed long-term
operating time (sometimes over 10 years). The PCL effect is relatively common in the VRLA batteries
of AGM type [9–17]. Currently, the problem has been reduced, but as a consequence, very often that
the batteries have to be exchanged more frequently, especially in key role systems. Three main models
of the process leading to the aforementioned phenomenon have been suggested:

• formation of a low-conducting layer at the grid/active mass interface (PCL of type 1). This is
probably the most frequent case of the PCL effect caused by the formation of a passivation layer
next to a typical corrosive layer at the grid/active mass interface. This tight coating contains
PbSO4 , and/or PbO1±n compounds,
• PCL of the second type is related to an increase in resistance of the pasted active mass (PAM),
which can be observed in case of weak adhesion between a grid and the PAM. The corrosive layer
consists mainly of β-PbO2 , and the concentration of sulphuric acid is far higher in this area. Thus,
the original spatial structure of the plate is distorted, which causes electrochemical insulation of a
part of the active mass. However, a decrease in active mass has not yet been observed.
• irreversible transformation of a superficial part of the active mass (usually negative) into big
lead sulphate crystals during cycling (PCL of the third type). Big crystals of PbSO4 are formed
gradually, and it is hard to reduce them into metallic Pb during charging [12–17].

In order to detect undergoing processes, the EIS technique seems to be very well suited. Huet [18]
analysed the relationship between various EIS parameters. He concluded that EIS can be an important
diagnostic tool. Application of a multi-channel EISmeter for precise measurements on batteries was
reported by Blanke et al. [19], who reviewed different approaches to determine SoC, SoH, and cranking
capability of lead–acid batteries with the aid of EIS.
Indeed, electrochemical impedance spectroscopy (EIS) is an excellent tool to analyze the
interfacial processes, variation in the internal resistance, state-of-charge, and the residual capacity
of a lead–acid battery. This method is fast, accurate, nondestructive in nature, and is ideal for
the modeling and diagnosis of industrial batteries. It was experimentally confirmed that this
technique can be applied to modeling battery behavior, as well as to the determination of the
state-of-charge (SoC) and the state-of-health (SoH) [20]. Diard et al. [21] conducted a general EIS
study on electrochemical battery discharge at a constant load and constant current; attention was
concentrated on the impedance test for a lead–acid battery with high capacity and low impedance
at load. The SoH is connected mostly with the ohmic resistance, the charge transfer resistance Rct ,
and the parameters of the constant phase element accounting for the diffusion in the pores when
the plate is well charged. In turn, battery modeling is an important issue for the design of different
battery applications. Thus, it is necessary to develop a reliable and easy-to-parameterize battery
model. However, for lead–acid batteries, impedance spectra depend very much on the short-term
charge-discharge history. Karden et al. [20,22] discussed further methodological questions connected
with the measurement and interpretation of the impedance spectra of batteries, especially nonlinearity,
voltage drift, stability, reproducibility, half-cell measurements, model structure, and parameter
extraction from the impedance data.
An alternative approach was suggested by Wei et al. [23,24] who assessed the battery state by
adopting a first-order RC model and proposing a dully-decoupled procedure which led to joint SoC and
capacity loss (SoH) estimation. This approach was later implemented in an online-monitoring method
to detect capacity loss in vanadium redox flow batteries in real time. It was checked experimentally
on the laboratory scale battery system and the SoC and the capacity loss was found to be accurately
derived in real time.
 Batteries 2018, 4, x 3 of 13
Batteries 2018, 4, 70 3 of 13

battery. The correlation between these factors would help to determine, as quickly as possible, the
initial
Thebattery capacity
aim of loss, which
this study can be
is to find regarded asbetween
correlations the beginning
voltageof changes,
the premature capacity
internal loss
resistance,
(PCL) effect, while
characteristics the system would
of electrochemical still havespectra,
impedance the capacity to be used
and actual safely.
capacity of the stationary lead-acid
In The
battery. the present work,
correlation we describe
between the preliminary
these factors would help results of these as
to determine, investigations. The study
quickly as possible, was
the initial
undertaken
battery because
capacity loss, of the low
which can effectiveness
be regarded as and reliability
the beginning of of
existing diagnostic
the premature systemsloss
capacity used in
(PCL)
emergency
effect, whilepower supply
the system modules.
would still have the capacity to be used safely.
In the present work, we describe the preliminary results of these investigations. The study was
2. Results and
undertaken Discussion
because of the low effectiveness and reliability of existing diagnostic systems used in
emergency
Duringpower
tests,supply modules.
an observed decrease in capacity of a 48 V battery that had operated in an
emergency power supply systems was nearly 30 Ah. The capacity after the first, second, and third
2.cycle
Results
was and
103 Discussion
Ah, 82 Ah and 76 Ah, respectively. For the new 48 V battery, the capacity did not
change and tests,
During had aanconstant value
observed of circa
decrease in110 Ah (Table
capacity 1).V battery that had operated in an emergency
of a 48
power supply systems was nearly 30 Ah. The capacity after the first, second, and third cycle was
103 Ah, 82Table 1. Capacity
Ah and of a brand new
76 Ah, respectively. 48 the
For V battery
new 48and
V the used battery,
battery, after float
the capacity did charge tests. and had a
not change
constant value of circa 110 Ah (Table 1). of a Brand New 48 V
Capacity Capacity of a Used 48 V
Time of Float Charge (d)
Battery (Ah) Battery (Ah)
Table 1. Capacity of a brand new 48 V battery and the used battery, after float charge tests.
0* 109 109
33 113
Capacity of a Brand New 48 V Capacity of 103a Used 48 V
Time of Float Charge (d)
66 114 (Ah)
Battery 82 (Ah)
Battery
99 0 * 116109 76
109
33* 0 day–batteries capacity after delivery
113 to tests and after recharging.103
66 114 82
99
Figure 1a,b show 116 of a brand new 48 V battery
changes of internal resistance 76 and used ones
during float charges in* the
0 day–batteries capacity
first, second, andafter delivery
third stagetooftests and(during
tests after recharging.
33, 66 and 99 days).
It can be noted that the internal resistance of a 48 V system is slightly higher for the used battery
thanFigure
for the1a,b
newshow
one,changes of internal
and it has a higher resistance
amplitude of aafter
brand fullnew 48 V battery
charging. and used
However, ones during
the values and
character
float of changes
charges of this
in the first, parameter
second, do not
and third have
stage a correlation
of tests (during with
33, 66electrical capacity.
and 99 days).

Figure 1. Changes in the internal resistance of 48 V (a) new battery and (b) used battery during float
Figure 1. Changes in the internal resistance of 48 V (a) new battery and (b) used battery during float
charge tests.
charge tests.

It can be noted that the internal resistance of a 48 V system is slightly higher for the used battery
Figure 2a–f show changes of the potential and the internal resistance of two selected
than for the new one, and it has a higher amplitude after full charging. However, the values and
monoblocks with 6 cells after operation in emergency power supply systems, as well as two new
character of changes of this parameter do not have a correlation with electrical capacity.
ones after 33, 66, and 99 days of a float charge. Selected units of 12 V had measured values of
Figure 2a–f show changes of the potential and the internal resistance of two selected monoblocks
potential and internal resistances which were very different from those of brand new ones.
with 6 cells after operation in emergency power supply systems, as well as two new ones after 33, 66,
Recorded data: E = f(t) and Rint = f(t) indicated that the internal resistance changes were small for
and 99 days of a float charge. Selected units of 12 V had measured values of potential and internal
systems previously used in emergency power supply systems, as well as for the new ones. An
resistances which
average value wereinternal
of the very different
resistancefrom those of brand
(measurements newfrom
data ones.
5 days to the end of the test, for
Recorded data: E = f(t) and R int = f(t) indicated that the internal
completely charged batteries—only float current in a circuit) is presented resistance changes
in Table 2. were small for
systems previously used in emergency power supply systems, as well as for the new ones. An average
 Batteries 2018, 4, 70 4 of 13

value of the internal resistance (measurements data from 5 days to the end of the test, for completely
Batteries
charged 2018,batteries—only
4, x float current in a circuit) is presented in Table 2. 4 of 13

Rint[mΩ]

Rint[mΩ]

Rint[mΩ]

Figure 2. Potential and internal resistance during (a,b) 33 days, (c,d) 66 days, (e,f) 99 days float charge
Figure 2. Potential and internal resistance during (a,b) 33 days, (c,d) 66 days, (e,f) 99 days float charge
tests, –two selected brand new 12 V monoblocks (dotted line), and two selected 12 V used monoblocks
tests, –two selected brand new 12 V monoblocks (dotted line), and two selected 12 V used monoblocks
(solid line).
(solid line).
Table 2. Average values of internal resistance for selected brand new 12 V monoblock and used
Table 2. Average
monoblock values
during of internal
float charge tests.resistance for selected brand new 12 V monoblock and used
monoblock during float charge tests.
Test Brand New (1) (Ω) Brand New (2) (Ω) Used (1) (Ω) Used (2) (Ω)
Test
33 days Brand New (1) (Ω) Brand New
0.0026 0.0026(2) (Ω) Used0.0033 (1) (Ω) Used0.0029 (2) (Ω)
33 days
66 days 0.0026
0.0026 0.0026
0.0027 0.0033
0.0033 0.0029
0.0029
99 days
66 days 0.0026
0.0026 0.0027
0.0027 0.0031
0.0033 0.0029
0.0029
99 days 0.0026 0.0027 0.0031 0.0029
It is noticeable that after 33 days of float charge tests (the first stage of the test), bigger changes in
It is noticeable
potential in case ofthat after12
the new 33Vdays of float charge
monoblocks tests (theOn
were observed. first
onestage of the possible
hand, test), bigger changes
reason for this
inmay
potential
be anin case of theformation
additional new 12 Vprocess
monoblocks were observed.
of monoblocks includedOn onein 48hand, the possible
V referential reason
battery. Onforthe
this may
other be an
hand, additional
during formation
the third stage, aprocess of monoblocks
bigger difference betweenincluded
valuesin of48
theVpotential
referential battery.for
is visible On the
the other hand, during the third stage, a bigger difference between values of
12 V used units after they stopped working. The measurements of changes in resistance of these two the potential is visible
for the 12ofVbatteries
groups used units after that
showed theyfor
stopped
all threeworking. The33,
cycles, i.e., measurements
66 and 99 days, of the
changes in resistanceafter
12 V monoblocks of
these two groups of batteries showed that for all three cycles, i.e., 33, 66
service had higher resistance parameters (in average about 3.0 mΩ) than the new ones (in average and 99 days, the 12 V
monoblocks
about 2.6 mΩ), afterand
service had higher
differences resistance
between parameters
their values during (inmeasurements
average aboutwere 3.0 mΩ) than the new
insignificant.
ones (inThe average about figures
subsequent 2.6 mΩ), and differences
(Figure 3a–f) showbetween their values
scatter diagrams during
of the measurements
potential were
and the internal
insignificant.
resistance of each 12 V unit in 48 V batteries drawn at different time of flow current tests.
The subsequent
It can be noticed figures
that 12(Figure
V units3a–f)
after show
servicescatter
had a diagrams of theand
large potential potential
internaland the internal
resistant span on
resistance
each stage ofof
each
the 12 V unit inelectric
presented 48 V batteries drawn
tests. Higher at different
potential and lowtimeinternal
of flow resistance
current tests.of the units after
It can
service be noticed
suggest that 12 Vwith
“problems” units after service
obtaining had a large
a completely potentialstate.
discharged and internal resistant span on
each stage of the presented electric tests. Higher potential and low internal resistance of the units
after service suggest “problems” with obtaining a completely discharged state.
Batteries 2018, 4, 70 5 of 13
Batteries 2018, 4, x 5 of 13

Rint[mΩ] Rint[mΩ]

Rint[mΩ] Rint[mΩ]

Rint[mΩ] Rint[mΩ]

Figure 3. Scatter diagrams of potential (bar graph) and internal resistance (line graph) of single 12 V
Figure 3. Scatter diagrams of potential (bar graph) and internal resistance (line graph) of single 12 V
monoblocks before start (a,b), after 5 days (c,d) and in last days (e,f) of flow current tests in 48 V battery:
monoblocks before start (a,b), after 5 days (c,d) and in last days (e,f) of flow current tests in 48 V
brand new (left–a,c,e), and used (right-b,d,f).
battery: brand new (left–a,c,e), and used (right-b,d,f).
Correlation between internal resistance and potential (on the last day of flow current tests—in
Correlation between internal resistance and potential (on the last day of flow current tests - in
33th, 66th and 99th day) of two selected 12 V units after service (with the highest differences of internal
33th, 66th and 99th day) of two selected 12V units after service (with the highest differences of internal
resistance) and new ones (as referential group) presented in Figures 4 and 5 shows average values of
resistance) and new ones (as referential group) presented in Figure 4 and 5 shows average values of
the potential of the units and internal resistance on the last day of current flow experiment, and 2 h
the potential of the units and internal resistance on the last day of current flow experiment , and 2h
after the end of this test (48 V batteries were in open circuit state).
after the end of this test (48 V batteries were in open circuit state).
Differences in the function Rint = f (E) can be observed for the 12 V units after operation in
Differences in the function Rint = f (E) can be observed for the 12 V units after operation in
emergency power supply systems in comparison with new 12 V units. It can be noticed that the
emergency power supply systems in comparison with new 12 V units. It can be noticed that the new
new 12 V units (excluding the first stage of current flow test which is probably the continuation
12 V units (excluding the first stage of current flow test which is probably the continuation of
of physicochemical changes in the active mass occurred after formation process) presented almost
physicochemical changes in the active mass occurred after formation process) presented almost
identical correlation Rint = f (E) in the 66th and 99th days. The change of this parameter for 12 V units
identical correlation Rint = f (E) in the 66th and 99th days. The change of this parameter for 12 V units
after service is not unambiguous, but slight trends in increasing potentials and decreasing internal
after service is not unambiguous, but slight trends in increasing potentials and decreasing internal
resistant can be observed. In Figure 5, slight differences in the function Rint = f (E) can be observed for
resistant can be observed. In Figure 5, slight differences in the function Rint = f (E) can be observed for
both groups, i.e., in the current flow state and open circuit state (2 h after cutting off the current flow).
both groups, i.e., in the current flow state and open circuit state (2 h after cutting off the current flow).
Lower span of the potential and the internal resistance for new 12 V units can be noticed.
Lower span of the potential and the internal resistance for new 12 V units can be noticed.
 Batteries 2018, 4, x 6 of 13
Batteries 2018, 4, 70 6 of 13
Batteries 2018, 4, x 6 of 13
3.4

3.4

2.9
2.9

2.4
2.4 13.40 13.45 13.50 13.55 13.60 13.65 13.70 13.75
13.40 13.45 13.50 13.55 13.60 13.65 13.70 13.75

Figure 4. Rint = f (E) on the last day of flow current tests (on the 33th, 66th and 99th day); figure legend:
Figure 4. R
Figure 4.int
square, = f= (E)
Rtriangle–two
int onon
f (E) thethelastlast
dayday
selected of
newflow
of 12current
flow teststests
V current
monoblocks;(on the
(on 33th,
theand
plus 66th
33th, and 99th
66th day);
andtwo
asterisk: 99th figurefigure
day);
selected legend:
used 12 V
square, triangle–two
legend: square, selected
triangle–two new
selected12 V
new monoblocks;
12 V monoblocks;plus and
plus asterisk:
and two
asterisk: two
monoblocks points in: dotted line area: values after 33 days of testing, discontinuous line area: selected
selected used
used V
12 values
monoblocks
V after
monoblocks points in: in:
points dotted
dotted line area:
line area:values
values after
after 33 days
days of
of testing,
testing,discontinuous
discontinuous line
line area:
area:
66 days of testing, solid line area: values after 99 days of testing.
values after
values 6666
after days
daysofoftesting,
testing,solid
solidline
linearea:
area: values after 99
values after 99days
daysofoftesting.
testing.

3.4
3.4
Rint [mΩ]
Rint [mΩ]

2.92.9

2.42.4
12.8
12.813.0 13.213.213.413.413.613.6
13.0 13.8 13.8
14.0 14.0

E [V]
E [V]
Figure
5. 5.Ratio
Figure RatioRR
5. Ratio intR=intf(E) on the
= f(E) 99th99th
on the day day
and and
2 h after finishing
2 h after the flow
finishing the current tests (each
flow current testsvalue
(eachisvalue is
Figure int = f(E) on the 99th day and 2 h after finishing the flow current tests (each value
an average
is anan average
average for two
fortwo
for selected
twoselected 12 V monoblocks)
selected1212VVmonoblocks) and
monoblocks)and the equation
andthetheequationof the
equationofof trend
thethe line.
trend
trend figure
line.
line. legend:
figure
figure legend:
legend:
rhombus–new
rhombus–new
rhombus–new batteries;
batteries; batteries; circle–used
circle–used batteries.
circle–used
batteries. solid solidsolid
batteries. point–before
point–before cutting
point–before
cutting the the float
floatcutting
current, the current,
float current,
discontinuous
discontinuous
discontinuous
point: point:
2 h after cutting point:
the2 float
h2after cutting
h current.
after the float
cutting current.
the float current.

Another
Another
Another technique
technique used
used
technique totocheck
used check the battery’s
the
to check battery’s
the state-of-health
state-of-health
battery’s was
waspotentiostatic
state-of-health electrochemical
potentiostatic
was electrochemical
potentiostatic electrochemical
impedance
impedance
impedance spectroscopy
spectroscopy
spectroscopy (PEIS),
(PEIS), applied
applied
(PEIS), to 12 V monoblocks
to 12toV12monoblocks
applied V monoblocks under
under OCP conditions
OCPOCP
under conditions (after batteries
(after
conditions batteries
(after batteries
potential
potential stabilization–2
stabilization—2 h after
h after finishing
finishing current
current flow
flow test).
test).
potential stabilization–2 h after finishing current flow test).
Two
Two groups
groups consistingofofthree
consisting three1212VV units
units each
each containing
containing new
new(one
(onegroup)
group)and
and used
used(another
(another
Two groups consisting of three 12 V units each containing new (one group) and used (another
group) batteries were investigated by EIS method after 33, 66, and 99 days of current flow tests.
group) batteries
group) were
batteries investigated
were byby
investigated EIS
EISmethod
methodafter
after 33, 66,
66, and
and99
99days
daysofof current
current flow
flow tests.
tests. Figure
6a presents impedance data for a chosen used battery (No. 12) in the form of Nyquist plots, while
Figure 6b presents impedance data for a chosen new battery (No. 15).
Batteries 2018, 4, 70 7 of 13
Batteries 2018, 4, x 7 of 13

Figure
Figure6a
6apresents
presentsimpedance
impedancedata
datafor
foraachosen
chosenused
usedbattery
battery(No.
(No.12)
12)in
inthe
theform
formof
ofNyquist
Nyquistplots,
plots,
while
whileFigure
Figure6b
6bpresents
presentsimpedance
impedancedata
datafor
fora achosen
chosennew
newbattery
battery(No.
(No.15).
15).

a) b)
0.10 0.14
33 days 33 days
66 days 0.12
0.08 66 days
99 days 99 days
0.10
0.06
0.08

-Im(Z) [W]
-Im(Z) [W]

0.04 0.06

0.04
0.02
0.02

0.00
0.00 0.02 0.04 0.06 0.08 0.10 0.00
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
-0.02 -0.02
Re(Z) [W] Re(Z) [W]

Figure6.6.Impedance
Figure Impedance spectra
spectra in the form
form of
ofNyquist
Nyquistplots
plotsfor:
for:(a)(a)1212
VVmonoblock
monoblockof of
used battery,
used (b)
battery,
1212
(b) VVmonoblock
monoblock of of
newnewbattery.
battery.

The
Theimpedance
impedancespectraspectrashown
shownininFigure
Figure6a,b 6a,bdemonstrate
demonstrateaadifference
differencebetween
betweenold oldand
andnewnew
batteries. It is seen that the Nyquist plot obtained after 99 days of float charge tests
batteries. It is seen that the Nyquist plot obtained after 99 days of float charge tests for the used for the used battery
isbattery
shiftedistoshifted
the right in comparison
to the with the with
right in comparison samethe plotsame
obtained for the new
plot obtained for battery.
the new battery.
AAmore
morequantitative
quantitativedescription
descriptionofofthe theexperimental
experimentaldata datarequires
requiresthe thechoice
choiceofofthe
theproper
properEECEEC
(electrochemical
(electrochemicalequivalent
equivalentcircuit).
circuit).The
Thesetsetofofimpedance
impedancedata datasubmitted
submittedfor foraafitting
fittingprocedure
procedurewaswas
reduced
reducedtotofrequency
frequency range
rangefrom 1 kHz
from 1 kHz to 1toHz to maintain
1 Hz to maintainthe linearity of theofsystem’s
the linearity response
the system’s [25].
response
The best-suited
[25]. EEC forEEC
The best-suited reduced impedance
for reduced data is shown
impedance data inisFigure
shown7.in This is a part
Figure of the
7. This iscircuit
a partwhich
of the
was used
circuit in many
which wasworks
used inconcerning
many works EISconcerning
investigations of lead-acid batteries
EIS investigations [19,25,26].
of lead-acid The[19,25,26].
batteries galvanic
cell
Thefrom lead-acid
galvanic cell batteries can be fitted
from lead-acid well with
batteries can bethefitted
model, having
well withfollowing
the model, electrical
having properties:
following
electrical properties:
• inductance L0 attributed to electrode geometry and connections inside the cell,
•  inductance
resistance RHFL0 due
attributed
to the to electrode geometry
connections, and connections
the separator, inside
the electrolyte the cell, and the surface
resistivity
 coverage
resistance R HF due to the connections, the separator, the electrolyte resistivity and the surface
of the electrodes by crystalline lead sulphate,
coverage of the electrodes by crystalline lead sulphate,
• resistance R1 depending on the porosity of the electrodes,
 resistance R1 depending on the porosity of the electrodes,
• CPE1 responsible for heterogeneous phenomena.
 CPE1 responsible for heterogeneous phenomena.
The
Theconstant
constantphase
phaseelement
elementwas
wasused
usedbecause
becausethe
thesurface
surfaceofofthe
theelectrode
electrodeisisnever
neverideally
ideallyflat,
flat,
and
andinin
thethe
electrochemical measurements,
electrochemical this element
measurements, replacesreplaces
this element the ideal the
capacitor
ideal [25]. The impedance
capacitor [25]. The
ofimpedance
the CPE isof
described
the CPE by the equation:
is described by the equation:
11
Q 
ZQZ= n (1)
QQ((j
jv )) n
whereQQisisaapre-exponential
where pre-exponentialfactor,
factor,which
whichisisaafrequency-independent
frequency-independentparameter;
parameter;nnisisthe
theexponent,
exponent,
√
which
whichdefines
definesthe
thecharacter
characterofoffrequency dependence(−(−1
frequencydependence 1 ≤≤nn≤≤ 1);
1); j j=  −11 is
is the
theimaginary
imaginaryunit;
unit;
and v = 2 πf [rad·s−−11 ] is angular frequency. With n = 1, Q is an ideal capacitance, and has the unit of
and  = 2 πf[rad·s ] is angular frequency. With n = 1, Q is an ideal capacitance, and has the unit of
capacitance (F); in another case when n 6= 1, Q has the unit of S·snn.
capacitance (F); in another case when n  1, Q has the unit of S·s .
Batteries 2018, 4, 70 8 of 13
Batteries 2018, 4, x 8 of 13

Figure 7. EEC
Figure used
7. EEC forformodeling
used modelingof
of the impedancespectra
the impedance spectraof of
12 12 V monoblock.
V monoblock.

The Theresults of the

results experimental
of the experimental data fittingwith
data fitting with chosen
chosen EECEEC (Figure
(Figure 7) are7) are in
given given
Tablein3. Table
To 3.
show that all data are well described with accepted EEC, Nyquist
To show that all data are well described with accepted EEC, Nyquist and Bode plots for the and Bode plots for the
experimental
experimental andand simulated
simulated dataare
data areshown
shown in in Figure
Figure 8. 8.The
Theexperimental
experimental data gathered
data in Table
gathered 3
in Table 3
demonstrate that the resistance R HF increases after 99 days by almost one order in magnitude (from
demonstrate that the resistance RHF increases after 99 days by almost one order in magnitude (from
circa 3 mOhms to 20 mOhms) for used batteries. This resistance change may indicate the beginning
circa 3 mOhms to 20 mOhms) for used batteries. This resistance change may indicate the beginning
of the PCL effect in the battery. This result seems to be reliable, since simulations of EIS spectra with
of the PCL effect in the battery. This result seems to be reliable, since simulations of EIS spectra with
accepted EEC agree well with the experimental results (Figure 8).
accepted EEC agree
The idea of well with the experimental
the application results
of electrochemical (Figure 8).
impedance spectroscopy for determination of
The idea of the application of electrochemical impedance
SoH and SoC is not a new one. In the case of industrial batteries, spectroscopy
it was discussed,for determination
e.g., by Huet [18] of SoH
and SoCand Karden et. al. [20,22]. However, the main problem lies in the interpretation of obtained and
is not a new one. In the case of industrial batteries, it was discussed, e.g., by Huet [18]
Karden et. al. [20,22].
impedance spectra.However,
The choicethe of main problemcircuit
an equivalent lies inisthe
notinterpretation
always universal. of obtained impedance
In our case, an
increase
spectra. Theinchoice
the resistance
of an for the used batteries
equivalent circuit is was detected,
not alwayswhich is a very In
universal. encouraging
our case,result for
an increase
in thethis non-destructive
resistance for thetest. Unfortunately,
used batteries was it seems to be the
detected, whichonlyis
reliable
a very information
encouraging extracted
resultfromfor this
non-destructive test. Unfortunately, it seems to be the only reliable information extracteditfrom
impedance spectra which can help to detect PCL effect. We found that for the low frequency range,
is difficult to find the proper EEC. More work is needed to work out a general approach to solve this
impedance spectra which can help to detect PCL effect. We found that for the low frequency range,
problem.
it is difficult to find the proper EEC. More work is needed to work out a general approach to solve
this problem. Table 3. Obtained results for impedance spectra according to the fit of EEC shown in Figure 7.

No. of Battery–Days L0 (H) RHF (Ω) R1 (Ω) QO (S·sn) n

Table 3. Obtained results for impedance spectra according to the fit of EEC shown in Figure 7.
U9-33 2.83·10−7 0.00276 0.03350 22.27 0.75
No. of Battery–DaysU9-66 L0 (H) 3.02·10R −7
HF0.00275
(Ω) 0.02449
R1 (Ω) 18.38QO (S0.78
·sn ) n
U9-99 3.52· 10 −7 0.01950 0.01879 24.70 0.72
U9-33 2.83 × 10−7 0.00276 0.03350 22.27 0.75
U10-33 2.68·10−70.00275
0.00315 0.01724
0.02449 27.31 18.38
0.71
U9-66 3.02 × 10−7 0.78
U9-99 U10-663.52 × 10−72.89·10 0.01950
−7 0.00330 0.02405
0.01879 17.67 24.70
0.78 0.72
U10-33 U10-992.68 × 10−73.69·10 0.00315
−7 0.01370 0.02088
0.01724 18.62 27.31
0.78 0.71
U10-66 U12-332.89 × 10−72.89·10 0.00330
−7 0.00288 0.01870
0.02405 28.53 17.67
0.70 0.78
U10-99 U12-663.69 × 10−73.14·10−70.01370
0.00299 0.02335
0.02088 18.34 18.62
0.77 0.78
U12-33 U12-992.89 × 10−73.80·10−70.00288
0.01990 0.02210
0.01870 18.80 28.53
0.77 0.70
U12-66 N13-333.14 × 10−72.97·10−70.00299 0.02335 18.83 18.34
0.00270 0.01817 0.77 0.77
U12-99 3.80 × 10 −7 0.01990 0.02210 18.80 0.77
N13-66 3.06·10 −7 0.00258 0.01664 20.05 0.75
N13-33 2.97 × 10 −7 0.00270 0.01817 18.83 0.77
N13-99 2.94·10 −7 0.00267 0.02458 14.12 0.81
N13-66 3.06 × 10 −7 0.00258 0.01664 20.05 0.75
N14-33 3.27·10 −7 0.00258 0.02084 16.93 0.79
N13-99 2.94 × 10 −7 0.00267 0.02458 14.12 0.81
N14-66 3.25·10 −7 0.00254 0.01569 19.12 0.76
N14-33 3.27 × 10 −7 0.00258 0.02084 16.93 0.79
N14-99 2.51·10 −7 0.00260 0.02276 14.11 0.81
N14-66 3.25 × 10−7 0.00254 0.01569 19.12 0.76
N15-33 3.01·10 −7 0.00260 0.02014 17.36 0.79
N14-99 2.51 × 10−7 0.00260 0.02276 14.11 0.81
N15-66 3.48·10 −7 0.00256 0.01606 18.63 0.76
N15-33 3.01 × 10−7 0.00260 0.02014 17.36 0.79
N15-99 2.28·10−7 0.00261 0.02302 14.16 0.81
N15-66 3.48 × 10−7 0.00256 0.01606 18.63 0.76
N15-99 2.28 × 10−7 0.00261 0.02302 14.16 0.81
Batteries 2018, 4, 70 9 of 13
Batteries 2018, 4, x 9 of 13

a) b)
0.025 0.012
33 days 33 days
66 days 0.010 66 days
0.020
99 days 99 days
33 days - fitting 0.008 33 days - fitting
0.015 66 days - fitting 66 days - fitting
99 days - fitting

-Im(Z) [W]
-Im(Z) [W]

0.006 99 days - fitting

0.010
0.004

0.005
0.002

0.000
0.000
0.000 0.005 0.010 0.015 0.020 0.025
0.000 0.002 0.004 0.006 0.008 0.010 0.012
-0.005 -0.002
Re(Z) [W] Re(Z) [W]
c) d)
40 50
40
30
30
20
20
10 10
-Phase  [ ]

-Phase  [ ]

0 0
33 days 33 days
66 days -10 66 days
-10 99 days
99 days
-20 33 days - fitting
33 days - fitting
-20 66 days - fitting 66 days - fitting
-30
99 days - fitting 99 days - fitting
-30 -40
1 10 100 1 000 1 10 100 1 000
Frequency [Hz] Frequency [Hz]

Figure
Figure 8. Experimental
8. Experimental andand fitting
fitting data data
in thein theofform
form of Nyquist
Nyquist and Bodeand Bode
plots plots12for:
for: (a,c) (a,c) 12 V
V monoblock
monoblock of used battery, (b,d) 12 V monoblock
of used battery, (b,d) 12 V monoblock of new battery. of new battery.

3. 3.
Materials
Materialsand
andMethods
Methods
Analysis
Analysisofof many correlations between
many correlations betweenphysicochemical
physicochemical features,
features, exploitation
exploitation conditions,
conditions, and
and electrical
electrical properties
properties andand their
their influence
influence on on
thethe capacity
capacity indicated
indicated several
several problems:
problems:

• • Acquiring
Acquiringalready
alreadyused
usedbattery
batterywith
withaaknown
knownhistory,
history, in
in order
order to
to increase
increase the
the probability
probability of
detection
detection ofof thePCL
the PCLeffect,
effect,
• • The mildness of the initial physicochemicalchanges
The mildness of the initial physicochemical changesaccompanying
accompanyingthe thePCL
PCLeffect,
effect,
• • The choice of tests and the testing conditions,
The choice of tests and the testing conditions,
• The appropriate classification of these changes and their assignment to the area of the PCL
• The appropriate classification of these changes and their assignment to the area of the PCL effect
effect (or other).
(or other).
To reduce these problems, the experimental material consisted of two groups of 12 V batteries
To reduce these problems, the experimental material consisted of two groups of 12 V batteries
(hereafter called “monoblocks”, in order to distinguish them from the 48 V system) AGM-VRLA
(hereafter called “monoblocks”, in order to distinguish them from the 48 V system) AGM-VRLA
with capacity of 100 Ah (C10). The 6-cell monoblocks of the first group had operated in emergency
with capacity
power supply of systems
100 Ah (C a ).telecommunication
in 10 The 6-cell monoblocks
stationoffor
theover
first4group
years.had
Those operated
systemsinconsisted
emergency of
power supply systems in a telecommunication station for over 4 years. Those
two batteries, and each battery included four monoblocks with 6 cells in every monoblock (each systems consisted
ofmonoblock
two batteries,
hadand
12 each batterypotential).
V nominal included four
The monoblocks
system is verywith 6 cells in
popular in every
mobilemonoblock
telephony(each
base
monoblock
stations. The 6-cell monoblocks of the second group were brand new, and were made base
had 12 V nominal potential). The system is very popular in mobile telephony stations.
by the same
The 6-cell monoblocks
manufacturer. of thenew
The brand second group
pieces were
were brand
used new, and were
as referential ones.made by the same
Construction manufacturer.
of the cells from
The brand new pieces were used as referential ones. Construction of the cells from
both groups of 12 V monoblocks (e.g.,: both–type AGM-VRLA–Figures 9 and 10, number of positive both groups of 12 V
monoblocks
and negative(e.g., both–type
plates, and gridAGM-VRLA–Figures
construction almost 9 and 10, number of 11)
identical–Figure positive and negative
indicated plates,
that almost all
differences in electrical and electrochemical properties should be associated with the usage history.
Batteries 2018, 4, 70 10 of 13

and grid construction almost identical–Figure 11) indicated that almost all differences in electrical and
electrochemical properties should be associated with the usage history.
Batteries
Batteries 2018,
2018, 4,
4, xx 10
10 of
of 13
13

Figure
Figure 9.
Figure 9. Brand
9. Brand new
Brand new 666 cells
new cells 12
cells 12 V
12 V monoblock
V monoblock (above),
monoblock (above), used
(above), used (below).
used (below).
(below).

Figure
Figure 10.
10. Cells
Cells in
in brand
brand new
new monoblock
monoblock (above),
(above), and
and used
used (below).
(below).

Some
Some information
information about
about thethe 48
48 V battery’s
battery’s use from data storage of the supervision automatic
automatic
system (a driver-cylinder device of MCSU-SU2300-600 type) was obtained (Figure 12a,b).
The average
averagevoltage
voltage of the
of the packpack
duringduring exploitation
exploitation was53.7
was about about 53.7
V, and theV, and the
average average
temperature
temperature was circa 300 K (a room without air-condition). The span of battery
was circa 300 K (a room without air-condition). The span of battery temperature was between 286temperature was K
between
and 312 K.286
TheK and 312 K.
number The number
of alerts (currentofinterruptions
alerts (current
in interruptions in energy
energy network) during network) during
use was <20, and use
the
was
total<20,
timeand the total
of outages time
was of 8outages
circa was circamonoblocks
h. All recovered 8 h. All recovered monoblocks
were operational (nowere operational
worrying (no
signs such
worrying signs such as high potential fluctuation or temperature increasing, either
as high potential fluctuation or temperature increasing, either in remote systems or by supervision in remote
systems or byobserved).
service, were supervision service, were observed).
Batteries 2018, 4, 70 11 of 13
Batteries 2018, 4, x 11 of 13
Batteries 2018, 4, x 11 of 13

Figure11.
Figure 11. Grids
Grids in
in brand
brand new
new monoblocks
monoblocks (above),
(above),and
andused
used(below); (+)(+)
(below); leftleft
(-)(-)
right.
right.
Figure 11. Grids in brand new monoblocks (above), and used (below); (+) left (-) right.

a)
a) b)
b)

Figure12. Number of discharges (a) and temperature values (b) of tested 48 V battery during
Figure 12. Number
Figure12.
operation of discharges
inNumber of power
emergency (a) and
discharges
supply(a) temperature values (b)
and temperature
systems (*installation of tested
values
in second 48ofV
(b)half
of battery
tested
2008 48 during operation
V battery
year). during
in operation
emergencyinpower supply
emergency systems
power (* installation
supply in second half
systems (*installation in of 2008 year).
second half of 2008 year).
The research consisted of float charge tests of systems containing 48 V batteries of 100 Ah (the
The
The
batteriesresearch
research
after consisted
consisted
operation ofoffloat
float
in emergency charge
powertests
charge ofof
tests
supply systems
systems
systems, containing
containing
and the new 4848VV batteries
ones). A flowof
batteries 100
of 100
current Ah Ah(the
test (the
batteries
was after
performed
batteries operation in
at 54.0 V (13.5
after operation emergency
V/monoblock)
in emergency power
power supply
in cycles systems,
supplyofsystems,
33, 66 and and
and the
99the
days,new
new ones).
using A flow
laboratory
ones). current
power test
A flow current
test was
supply
was performed
RXN
performed (voltage at range
at 54.054.0 V0–60
V (13.5(13.5V,V/monoblock)
current range
V/monoblock) in cycles
0–5
in cycles of66
33,and
A).ofMeasurements
33, 66 99anddays,
of99 days,
usingusing
internal laboratory
resistance
laboratory and
power
power supply
voltage of bothRXNthe (voltage
battery range
and 0–60
each V, current
single range
monoblock 0–5
were A). Measurements
performed
supply RXN (voltage range 0–60 V, current range 0–5 A). Measurements of internal resistance and once aof internal
day, during resistance
float
and voltage
charge,
voltage of of
using both
bothHioki the
the battery
device
battery andeach
model
and eachsingle
3554. single
After 33,monoblock
66, and 99
monoblock were
days
were performed once
onceaathe
of float charge,
performed day, during
circuit
day, during float
was float
charge,
charge,using
unplugged. The
using Hioki device
batteries
Hioki model
were
device 3554.
left for
model After
2 h at
3554. open33,
After 33,66,66,and
current and9999days
circuit and
days ofof
the float
floatcharge,
charge,the
electrochemical thecircuit
impedance
circuitwaswas
unplugged.
spectroscopy The batteries
(EIS) were
measurement left for
of 2
singleh at
12open
V current
monoblock circuit
was and the
performed electrochemical
unplugged. The batteries were left for 2 h at open current circuit and the electrochemical impedance using a impedance
BioLogic CLB
2000 galvanostat-potentiostat
spectroscopy
spectroscopy (EIS) measurement
(EIS) measurement with of20single
of single A12current
V12 booster.was
monoblock
V monoblock EISperformed
wasexperiments using
performed were run
a BioLogic
using with
CLBAC
a BioLogic 2000
CLB
amplitude of 40 mV over the frequency range of 10 3  5 × 10−3 Hz (10 points per decade). Impedance
galvanostat-potentiostat
2000 galvanostat-potentiostat with 20 A current booster. EIS experiments were run with of
with 20 A current booster. EIS experiments were run with AC amplitude AC
data were analyzed using ZView2 software (Scribner Associates Inc., Southern Pines, NC, USA).
amplitude of 40 mV over the frequency range of 10 3  5 × 10−3 Hz (10 points per decade). Impedance
data were analyzed using ZView2 software (Scribner Associates Inc., Southern Pines, NC, USA).
Batteries 2018, 4, 70 12 of 13

40 mV over the frequency range of 103 ÷ 5 × 10−3 Hz (10 points per decade). Impedance data were
analyzed using ZView2 software (Scribner Associates Inc., Southern Pines, NC, USA).
Finally, the 48 V battery was discharged using Digatron BNT 50-100-3 (voltage range 5–100 V,
current range 0.05–50 A) by I10 current until the voltage of 43.2 V was reached. The aim of the
measurement was to assess changes in the capacity of a 48 V battery and/or single 12 V monoblocks,
and to link these changes with the aforementioned measurements. Next, the 48 V battery was plugged
again into laboratory power supply and a subsequent cycle of float charge started.

4. Conclusions
Two different approaches were implemented to determine the premature capacity loss effect in
batteries operating in backup power sources of telecommunication antenna stations. The first approach
consisted of taking measurements of voltage and resistance change of 12 V old and new monoblocks
as a function of time. Slight changes in voltage and resistance were recorded; however, no visible trend
was found which could indicate the beginning of a PCL effect. In turn, electrochemical impedance
spectroscopy enabled us to detect resistance changes which could be attributed to the beginning of
the PCL effect. This method is promising, but its success depends on the choice of equivalent circuit,
which is not universal. It seems that the choice of the frequency range during experiments is crucial to
obtaining reliable results. One additional advantage of EIS technique is its non-disruptive application
i.e., the backup power system does not need to be switched off before the test, as it is in the case of
direct capacity measurement tests.

Author Contributions: W.M.—participation in the development of the project concept, participation in the
analysis and development of results, management and content-related supervision. E.J.—sample preparation for
measurements, performance of experiments, participation in the analysis and development of results, responsible
for comments in the publication. M.B.—participation in the development of the project concept, participation in
the analysis and development of results, development of research methods, responsible for writing the paper.
P.H.—participation in the development of the project concept, participation in the analysis and development
of results, development of research methods, responsible for writing the paper. R.S.—participation in the
development of the project concept, participation in the analysis and development of results.
Funding: The National Centre for Research and Development, Poland: PBS1/B4/3/2012.
Acknowledgments: Research carried out by the project financed by the Polish National Centre for Research and
Development (PBS1/B4/3/2012—“Diagnosis of early recognition of the PCL-phenomenon in lead-acid batteries
to increase the reliability of emergency power supply systems”). Authors are grateful to Ph.D. Michał Stepień for
his assistance in impedance data analysis and to Prof. Krzysztof Fitzner for valuable discussion.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Sato, S.; Kawamura, A. A new estimation method of state of charge using terminal voltage and internal
resistance for lead acid battery. In Proceedings of the Power Conversion Conference, Osaka, Japan,
2–5 April 2002; Volume 2, pp. 565–570.
2. Zhendong, Z.; Zihan, L.; Nan, H. The VRLA battery internal resistance on-line measuring device.
In Proceedings of the 2012 International Conference on Information Management, Innovation Management
and Industrial Engineering (ICIII), Sanya, China, 20–21 October 2012; pp. 404–406.
3. Baraniak, M.; Jankowska, E.; Budzyński, M.; Lula, M.; Majchrzycki, W. Report IMN-CLAiO No. 6093.01 in
Polish; Institute of Non-Ferrous Metals: Poznań, Poland, 2012.
4. Majchrzycki, W.; Jankowska, E.; Baraniak, M. Report IMN-CLAiO No. 6436.01 in Polish; Institute of
Non-Ferrous Metals: Poznań, Poland, 2013.
5. Abraham, P.; Bača, P.; Vaculík, S. Resistivity and Impedance Changes of Lead-Acid Accumulator. ECS Trans.
2014, 48, 303–308. [CrossRef]
6. Gençten, M.; Dönmez, K.B.; Şahin, Y.; Pekmez, K.; Suvac, E. Voltammetric and electrochemical impedimetric
behavior of silica-based gel electrolyte for valve-regulated lead-acid battery. J. Solid State Electrochem. 2014,
18, 2469–2479. [CrossRef]
Batteries 2018, 4, 70 13 of 13

7. Shih, H.; Lo, T.-C. Electrochemical Impedance Spectroscopy for Battery Research and Development Technical Report
No 31; Solartron: Farnborough, UK, 1996.
8. Fasih, A. Modeling and Fault Diagnosis of Automotive Lead-Acid Batteries. Master’s Thesis, The Ohio State
University, Columbus, OH, USA, 2 April 2006.
9. Moseley, P.T.; Garche, J.; Parker, C.D.; Rand, D.A.J. Valve-Regulated Lead-Acid Batteries, 1st ed.; Elsevier Science:
Amsterdam, The Netherlands, 2004; ISBN 9780080474731.
10. Pavlov, D. Lead-Acid Batteries: Science and Technology, 2nd ed.; Elsevier Science: Amsterdam, The Netherlands,
2017; ISBN 9780444595607.
11. Ł˛egosz, Z. Czy baterie VRLA moga˛ żyć 20 lat? Przedwczesna utrata pojemności—Fundamentalny problem
baterii VRLA. Wiadomości Elektrotechniczne 2001, R. LXIX, 111–115.
12. Calábek, M.; Micka, K.; Bača, P.; Křivák, P.; Šmarda, V. Resistance changes and premature capacity loss in
lead battery plates. J. Power Sources 1996, 62, 161–166. [CrossRef]
13. Hollenkamp, A.F. Premature capacity loss in lead/acid batteries: A discussion of the antimony-free effect
and related phenomena. J. Power Sources 1991, 36, 567–585. [CrossRef]
14. Kosai, M.; Yasukawa, S.; Osumi, S.; Tsubota, M. Effect of antimony on premature capacity loss of lead/acid
batteries. J. Power Sources 1997, 67, 43–48. [CrossRef]
15. Pavlov, D. Effect of corrosion layer on phenomena that cause premature capacity loss in lead/acid batteries.
J. Power Sources 1994, 48, 179–193. [CrossRef]
16. Pavlov, D.; Petkova, G.; Dimitrov, M.; Shiomi, M.; Tsubota, M. Influence of fast charge on the life cycle of
positive lead–acid battery plates. J. Power Sources 2000, 87, 39–56. [CrossRef]
17. Shiomi, M.; Okada, Y.; Tsuboi, Y.; Osumi, S.; Tsubota, M. Study of PCL mechanism: Influence of grid/PAM
state on PCL. J. Power Sources 2003, 113, 271–276. [CrossRef]
18. Huet, F. A review of impedance measurements for determination of the state-of-charge or state-of-health of
secondary batteries. J. Power Sources 1998, 70, 59–69. [CrossRef]
19. Blanke, H.; Bohlen, O.; Buller, S.; De Doncker, R.W.; Fricke, B.; Hammouche, A.; Linzen, D.; Thele, M.;
Uwe Sauer, D. Impedance measurements on lead–acid batteries for state-of-charge, state-of-health and
cranking capability prognosis in electric and hybrid electric vehicles. J. Power Sources 2005, 144, 418–425.
[CrossRef]
20. Karden, E.; Buller, S.; De Doncker, R.W. A method for measurement and interpretation of impedance spectra
for industrial batteries. J. Power Sources 2000, 85, 72–78. [CrossRef]
21. Diard, J.-P.; Le Gorrec, B.; Montella, C. EIS study of electrochemical battery discharge on constant load.
J. Power Sources 1998, 70, 78–84. [CrossRef]
22. Karden, E.; Buller, S.; De Doncker, R.W. A frequency-domain approach to dynamical modeling of
electrochemical power sources. Electrochim. Acta 2002, 47, 2347–2356. [CrossRef]
23. Wei, Z.; Tseng, K.J.; Wai, N.; Lim, T.M.; Skyllas-Kazacos, M. Adaptive estimation of state of charge and
capacity with online identified battery model for vanadium redox flow battery. J. Power Sources 2016, 332,
389–398. [CrossRef]
24. Wei, Z.; Bhattarai, A.; Zou, C.; Meng, S.; Lim, T.M.; Skyllas-Kazacos, M. Real-time monitoring of capacity
loss for vanadium redox flow battery. J. Power Sources 2018, 390, 261–269. [CrossRef]
25. Barsoukov, E.; Macdonald, J.R. (Eds.) Impedance Spectroscopy Theory, Experiment, and Applications, 2nd ed.;
John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2005.
26. Zhu, W.H.; Zhu, Y.; Tatarchuk, B.J. A simplified equivalent circuit model for simulation of Pb–acid batteries
at load for energy storage application. Energy Convers. Manag. 2011, 52, 2794–2799. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).